While I can’t put my raw spreadsheet up online for a number of reasons, I have had requests to make the data somehow available. I’ve thus compiled a simplified version of my spreadsheet (TV anime only). Note, I do not plan to keep this updated because it’s something of a pain to reformat the data like this.

As usual, if you notice anything incorrect (release dates and seasons/years in particular) please let me know so I can correct it.

Comments
This spreadsheet contains the following columns:
A. Title
B. Volume number
C. Disc format
D-M. Weekly data, 1 through 10
N. Additional data
O. Total (per format)
P. Total (all formats, per volume)
Q. Percentage drop from previous volume
R. Percentage drop from first volume
S. Release date
T. Airing season
U. Airing year
V. Total number of volumes
W. Stalker estimate in points
X. Stalker estimate accuracy
Y. MSRP with tax
Z. MSRP without tax
AA. Total sales in yen (MSRP without tax * units sold)
AB. Source material type
AC. Animation production studio
AD. Disc publisher (gets paid for each unit shipped)
AE. Disc distributor (gets paid a fee for distribution work)

A comment on some of the cell coloring:

Formatting example

1. Purple indicates data that came from the extended rankings. Note that in some cases multiple weeks of data are included in the last cell. This would happen, for example, in a case where a title had ranked three times before we had extended rankings, and then upon checking the extendeds, it was recorded with 5+ weeks. Since I can’t tell which of that data was wk4 vs wk5 vs wk6, I just put the sum in one cell. You’ll also sometimes see blank purple cells. These are situations where Oricon indicates a disc ranked again, but the ranking wasn’t enough to hit the next milestone and increase the sales number.

2. Yellow indicates data that came from anything else other than the normal weekly rankings. Primarily this is from previous unranked weeks getting filled in by a later week in which the release reappears, though there’s a smattering of the old Anime Power Rankings and stuff in there.

3. Green indicates it’s a value from the Remainder column. This typically means data from weeks 11 onward, though it can also be monthly, mid-year, or yearly data that couldn’t be tied to specific weeks. In a few cases, like Macross Frontier, we had no weekly ranking data at all in the early days of BD rankings, and the sales we know only come from the yearly rankings.

Formatting example 2

4. This bluish color in the Stalker estimates column indicates a release on 2015-09-25/26. Those were affected by Oricon not reporting Amazon data, and thus the “Accuracy” column is heavily skewed and makes the estimate look worse than it really was.

Download
Anime TV Sales, simplified full data dump, Updated 2015-10-24
* If you’re doing anything other than quickly glancing at it, I recommend downloading it to Excel (File > Download as) before trying to do anything substantial with it.

2 Responses to “Anime TV Sales, simplified full data dump, Updated 2015-10-24”

  1. primadog says:

    If I spent thousands of hours on a project like this, I’d have reservations about just putting it out in public. So thank you, again, for making these stats available to us wonks.

    The Stalker Accuracy column catched my eyes this time. For a long time I assume that the Stalker algorithm are becoming increasingly more accurate over the years. Surprising, your data doesn’t back that intuition. Standard deviation is quite high, even though it’s centered slightly above zero. In other words, Stalker is pretty accurate, but not exactly precise.

    What pops out though, is the scatter has actually grown the past couple years (STDEV 2011: 32% -> 2015: 45%). I guess it has to do with the growth of fujoshi market?

    http://imgur.com/a/ngIQ7

    • something
      something says:

      I’d guess what’s hurting killing Stalker recently is the huge increase in Amazon-exclusive editions. Those would be a nightmare to adjust for. I think the problem there is that the algorithm for a normal disc is built around the correct assumption that Amazon is not 100% of the market, but for Amazon-exclusive editions they *are* 100% of the market by definition. But those Amazon editions get the same points per rank as a disc that could also be selling at thousands of other retailers.

      But I haven’t analyzed the data to see if Amazon editions suffer from this more often.

Leave a Reply